On Sun, May 22, 2022 at 09:55:58PM -0700, Sean Li wrote: > Please refer to the sniffer capture at > https://drive.google.com/file/d/13EcsEJKjKrkID20SPaq0YYA3_iatGD6Y/view?usp=sharing > Pixel6 mac: 0c:c4:13:14:16:93 > The issue usually happens after multiple roaming connections between > our mesh system. Thanks. This is interesting.. The station tries to reassociate without H2E and once that fails, associates again with H2E with the same AP. > If the 802.11 spec mandates H2E when both peers advertised the > support, the issue here sounds more on the Pixel6 then. That's what it look like, but the that sniffer capture has some details that made me look at what else could be wrong. It looks like the station might be doing this particular roaming case based on FILS Discovery frame and that frame does not have information about SAE H2E, so the station might not have figured out that the target AP supports H2E. That said, this is on the 6 GHz band, so the AP should be assumed to use H2E anyway. Furthermore, if the station does not have the sufficient information from the FILS Discovery frame, it needs to wait to receive the full information from Beacon frame or Probe Response frame before trying to association.. I'm not familiar with the particular station implementation here for the relevant parts, so I'll talk with the vendor in question to check what could be behind this. In addition, I went through the FILS Discovery frame details and compared it to the way this frame is defined in the standard. I don't think I agree with the Length field use here, so that's something that I'll take a closer look as well. hostapd has similar design in setting the Length field value to point beyond the TPE element, so there might be multiple implementations doing this differently compared to my interpretation. I guess this requires some more discussions to confirm my interpretation. Anyway, the frame can still be parsed successfully, so I don't think this would have anything to do with that SAE H2E issue. -- Jouni Malinen PGP id EFC895FA _______________________________________________ Hostap mailing list Hostap@xxxxxxxxxxxxxxxxxxx http://lists.infradead.org/mailman/listinfo/hostap
